Abstract
Background:
Housing status impacts outcomes after elective and emergent operations but has not been well studied in the emergency general surgery population. This study investigates the impact of housing status on complications and 30-day follow-up, emergency department visits, and readmissions after emergency general surgery admission.
Methods:
We conducted a retrospective matched cohort study of adult patients admitted with an emergency general surgery diagnosis at an urban, safety net hospital from 2014 to 2021. Patients were matched 1 to 2 on the basis of age, sex, Charlson Comorbidity Index, diagnosis, and operative status. The primary exposure was unhoused status. The primary outcome was in-hospital complications. Secondary outcomes included intensive care unit admission, extended length of stay, follow-up attendance, and emergency department visit or unplanned readmission within 30 days. Multivariable conditional logistic regression was used to determine the association between housing status and the outcomes of interest.
Results:
The study included 531 patients (177 unhoused, 354 housed). There were no significant differences in complications, intensive care unit admissions, or extended length of stay. Unhoused patients had lower odds of outpatient follow-up (odds ratio, 0.54; 95% confidence interval, 0.35–0.85, P = .008) and higher odds of emergency department utilization (odds ratio, 2.72; 95% confidence interval, 1.78–4.14, P < .001) and readmission (odds ratio, 1.87; 95% confidence interval, 1.09–3.19, P = .02).
Conclusion:
Compared with housed patients, unhoused patients with emergency general surgery conditions have lower rates of outpatient follow-up and greater odds of using the emergency department and being readmitted within 30 days of discharge. This points to a need for dedicated posthospitalization care and creative methods of engaging with this population.
Introduction
Housing insecurity impacts more than half a million individuals in the United States, largely because of a lack of affordable housing and public aid.1,2 Unhoused individuals experience greater rates of acute health care needs, including surgical care.3–5
As a result of the challenges of obtaining routine outpatient care within the community, unhoused patients are more likely than housed patients to access the emergency department.6,7 Furthermore, unhoused patients are more likely to present at the extremes of illness, often presenting to the emergency department with either nonurgent complaints that do not require hospital admission or with advanced or severe disease.7–9 The epidemiology of unhoused patients presenting to the emergency department has been well described, with individuals experiencing homelessness more likely to be young, male, uninsured, socially isolated, and have mental health comorbidities.10,11 For the subset of these patients who require surgical evaluation, unhoused status presents barriers at all phases of care, which may impact outcomes.12 Previous literature has demonstrated that surgical care differs between housed and unhoused patients, paralleling the impact of housing status on outcomes in other aspects of health care; however, these studies have not evaluated patients who undergo emergency general surgery specifically.12,13 Furthermore, patients who undergo emergency general surgery have high rates of housing insecurity, which may distinguish this subset from the overall surgical patient population.14
Given the sparse literature evaluating outcomes of unhoused patients requiring admission for emergency general surgery pathologies, there is a need for more specific evaluation of this population.15 To our knowledge, there are no matched studies comparing the outcomes of emergency general surgery diagnoses between housed and unhoused patients. Understanding in-hospital and postdischarge outcomes in these patients will allow support for future targeted interventions to improve care for this population. In this study, we evaluate the impact of housing status on in-hospital (complications, intensive care unit [ICU] admission, extended length of stay), and 30-day posthospitalization (follow-up attendance, emergency department visit, unplanned readmission) outcomes after emergency general surgery admission in a tertiary care, safety net hospital. We hypothesized that unhoused patients would experience greater rates of in-hospital complications, lower rates of follow-up attendance, and greater rates of emergency department use and unplanned readmission.
Methods
Data source and study participants
We performed a retrospective, matched cohort study of unhoused adults between the ages of 18 and 89 years admitted to our urban, safety net hospital with 1 of 11 emergency general surgery diagnoses between June 1, 2014, and December 1, 2021. Data were obtained from our institution’s clinical data warehouse, followed by manual chart review to validate the accuracy of included data. Potentially eligible patients were identified for study inclusion based on administrative billing codes, specifically International Classification of Diseases, Ninth or Tenth Revision codes, corresponding to select diagnosis groups, including acute uncomplicated appendicitis, acute complicated appendicitis, acute cholecystitis, small bowel obstruction, incarcerated or strangulated inguinal hernia, incarcerated or strangulated umbilical or ventral hernia, uncomplicated colonic diverticulitis, complicated colonic diverticulitis, gastric, duodenal or unspecified peptic ulcer with perforation, or perirectal abscess (Supplementary Table S1). For individuals with multiple admissions for one of the included diagnosis codes, only the first admission was used. After initial identification of patients for inclusion, manual chart review was completed by trained study personnel to confirm the diagnosis and that it was the principal reason for hospitalization.
This study was deemed exempt by our institutional review board, and a waiver of informed consent was granted. Study data were collected and managed using Research Electronic Data Capture, hosted at Boston University.16,17 The EQUATOR Network Strengthening the Reporting of Observational Studies in Epidemiology guideline was used to ensure the proper reporting of data and results (Supplementary Table S2).
Matching
To control for potential confounders, housed and unhoused patients were matched 2:1 on the basis of age (within 5 years), sex, Charlson Comorbidity Index (within 3 points), diagnosis group as described previously, and dichotomous operative status. If housing status, diagnosis group, or surgical intervention was determined by chart review to be incorrect for a patient recorded as unhoused, both that patient and their 2 corresponding control (housed) patients were excluded from analysis. If these factors were determined to be incorrect for a control (housed) patient, an alternate control (with the same study eligibility and matching criteria) was identified and included in the study. If both controls were excluded or a satisfactory second control could not be identified, the corresponding case was excluded.
Primary exposure
The primary exposure was unhoused status. A stepwise method was used to screen for eligible unhoused patients. First, patients were screened positive for unhoused status if they met any of the following criteria during their index admission, as determined by electronic medical record data recorded in our clinical data warehouse: hospital intake form with positive screening question for housing insecurity or homelessness, patient-reported unhoused status during hospital registration, patient address corresponding to a shelter, or record of homelessness on the problem list in the electronic health record. Next, housing status was confirmed by trained study personnel via manual chart review, evaluating for any documentation of unhoused status within the year leading up to the date of hospital admission for an emergency general surgery diagnosis.
Primary outcome
The primary outcome was in-hospital complications or adverse events. Adverse events were defined in accordance with the American College of Surgeons National Surgical Quality Improvement Program. Complications and adverse events occurring during the index admission were classified into the following groups: wound disruption (including dehiscence or evisceration), superficial surgical-site infection, deep surgical-site infection, organ space surgical site infection, urinary tract infection, pneumonia, sepsis or septic shock, unplanned intubation, ventilator support for >48 hours, stroke, myocardial infarction, cardiac arrest, acute renal failure, deep venous thrombosis, pulmonary embolism, hemorrhage requiring transfusion of blood products, and death.
Secondary outcomes
Secondary outcomes were ICU admission, extended hospital length of stay, attendance at clinic follow-up within 30 days of hospital discharge, and an emergency department visit or unplanned readmission within 30 days of discharge. Extended length of stay was defined as greater than the 75th percentile length of stay for all patients (4.8 days), similar to previous literature.18 Outpatient clinic follow-up within 30 days was included only if it was related to the index hospitalization, as appointments unrelated to the hospitalization were deemed to represent unique instances of care rather than follow-up. Follow-up did not have to be with a surgeon to be included. Follow-up visit types were classified as primary care or surgical/procedural specialties. All emergency department visits and unplanned readmissions within 30 days of index admission were identified. Primary reason for an emergency department visit or unplanned readmission was evaluated to designate whether the reason for re-presentation was related to the index admission.
Covariates
Demographic information included age, sex, race (classified as White, Black, or other), ethnicity (Hispanic/Latino or not Hispanic/Latino), primary language (classified as English, Spanish, or other), insurance status (classified as commercial, Medicaid, Medicare, or other), and Charlson Comorbidity Index. Hospitalization information included type of admission (observation versus inpatient), undergoing an operation or non-surgical procedure, return to the operating room during index admission, length of stay, and discharge disposition. Operations were classified both dichotomously and by type. Disposition was categorized as home (including home with home healthcare services), rehabilitation, skilled nursing facility, long-term acute care facility, psychiatric facility, shelter or respite facility, against medical advice, and other.
Missing data
Variables with missing data included race (n = 43, 8.10%), ethnicity (n = 4, 0.75%), and insurance status (n = 248, 46.70%) (Table I). Data for all variables used in adjusted analyses were complete, with the exception of insurance status. Patients with missing insurance status were designated as not having commercial insurance for the purpose of adjusted analysis.
Table I.
Participant demographics and presenting characteristics
| Characteristics | All patients | Unhoused | Housed | P value |
|---|---|---|---|---|
| (N = 531) | (n = 177) | (n = 354) | ||
| Male, n (%) | 351 (66.10) | 117 (66.10) | 234 (66.10) | – |
| Age, yr, median [IQR] | 51 [38–59] | 51 [38–59] | 51 [38–60] | – |
| Charlson Comorbidity Index, median [IQR] | 1 [0–3] | 1 [0–3] | 1 [0–3] | – |
| Surgery during admission, n (%) | 267 (50.28) | 89 (50.28) | 178 (50.28) | – |
| Diagnosis group, n (%) | ||||
| Acute uncomplicated appendicitis | 78 (14.69) | 26 (14.69) | 52 (14.69) | – |
| Acute complicated appendicitis | 9 (1.69) | 3 (1.69) | 6 (1.69) | |
| Acute cholecystitis | 150 (28.25) | 50 (28.25) | 100 (28.25) | |
| Acute pancreatitis | 144 (27.12) | 48 (27.12) | 96 (27.12) | |
| Small bowel obstruction | 30 (5.65) | 10 (5.65) | 20 (5.65) | |
| Complicated colonic diverticulitis | 6 (1.13) | 2 (1.13) | 4 (1.13) | |
| Uncomplicated colonic diverticulitis | 48 (9.04) | 16 (9.04) | 32 (9.04) | |
| Incarcerated or strangulated inguinal hernia | 15 (2.82) | 5 (2.82) | 10 (2.82) | |
| Incarcerated or strangulated umbilical or ventral hernia | 33 (6.21) | 11 (6.21) | 22 (6.21) | |
| Peri-rectal abscess | 18 (3.39) | 6 (3.39) | 12 (3.39) | |
| Race, n (%) | ||||
| White | 144 (27.12) | 54 (30.51) | 90 (25.42) | .12 |
| Black | 265 (49.91) | 83 (46.89) | 182 (51.41) | |
| Other | 79 (14.88) | 31 (17.51) | 48 (13.56) | |
| Declined/not available | 43 (8.10) | 9 (5.08) | 34 (9.60) | |
| Hispanic or Latino, n (%) | ||||
| Yes | 113 (21.28) | 41 (23.16) | 72 (20.34) | .72 |
| No | 414 (77.97) | 135 (76.27) | 279 (78.81) | |
| Unknown or not reported | 4 (0.75) | 1 (0.56) | 3 (0.85) | |
| Primary language, n (%) | ||||
| English | 454 (85.50) | 153 (86.44) | 301 (85.03) | .19 |
| Spanish | 57 (10.73) | 21 (11.86) | 36 (10.17) | |
| Other | 20 (3.77) | 3 (1.69) | 17 (4.80) | |
| Insurance, n (%) | ||||
| Commercial | 54 (10.17) | 6 (3.39) | 48 (13.56) | <.001 |
| Medicaid | 188 (35.40) | 82 (46.33) | 106 (29.94) | |
| Medicare | 39 (7.34) | 15 (8.47) | 24 (6.78) | |
| Other | 2 (0.38) | 0 (0.00) | 2 (0.56) | |
| Missing or not reported | 248 (46.70) | 74 (41.81) | 174 (49.15) | |
| Hospital encounter, n (%) | ||||
| Inpatient admission | 368 (69.30) | 124 (70.06) | 244 (68.93) | .79 |
| Observation | 163 (30.70) | 53 (29.94) | 110 (31.07) |
IQR, interquartile range.
Statistical analysis
After matching, sociodemographic, hospitalization, and outcome variables were compared between housed and unhoused individuals. Patient characteristics were compared using the χ2 test for categorical variables, and Wilcoxonrank sum test for continuous variables, which were nonparametric. Next, multivariable conditional logistic regression was conducted to further evaluate the impact of our variables of interest on our primary and secondary outcomes, controlling for age, sex, primary language, insurance status, housing status, Charlson Comorbidity Index, diagnosis group, surgical intervention, and ICU admission. We chose to exclude race, a frequently adjusted for social determinant of health, from our adjusted analysis as it represents as construct rather than a true risk factor for adverse outcomes. In addition, as race has been shown to be significantly related to insurance status,19 we felt that using insurance status was a more appropriate reflection of disparity. In order to evaluate the impact of discharge disposition on readmissions, a sensitivity analysis was conducted comparing patients discharged home to those discharged to any nonhome destination. A sensitivity analysis was then conducted including only patients who underwent an operation. Statistical analysis was completed using STATA, version 15.1 (StataCorp LLC, College Station, TX).
Results
Patient characteristics
A total of 531 patients were identified for study inclusion (Figure 1). The study cohort had a median age of 51 years (interquartile range, 38–59), was 66.10% male (n = 351), and 49.91% (n = 265) Black (Table I). Overall, 267 patients (50.28%) underwent an operation, most commonly laparoscopic cholecystectomy (n = 116, 43.45%) or laparoscopic appendectomy (n = 76, 28.46%) (Table II).
Figure 1.
Study cohort and exclusion consort diagram indicating inclusion and exclusion criteria and each stage of screening.
Table II.
Interventions and in-hospital characteristics
| Characteristics | All patients | Unhoused | Housed | P value |
|---|---|---|---|---|
| (N = 531) | (n = 177) | (n = 354) | ||
| Surgery during admission, n (%) | 267 (50.28) | 89 (50.28) | 178 (50.28) | – |
| Surgical intervention, n (%) | ||||
| Laparoscopic appendectomy | 76 (28.46) | 26 (29.21) | 50 (28.09) | .73 |
| Laparoscopic cholecystectomy | 116 (43.45) | 39 (43.82) | 77 (43.26) | |
| Partial or total colectomy | 2 (0.75) | 1 (1.12) | 1 (0.56) | |
| Small bowel resection | 2 (0.75) | 1 (1.12) | 1 (0.56) | |
| Laparoscopy with lysis of adhesions and washout | 6 (2.25) | 1 (1.12) | 5 (2.81) | |
| Exploratory laparotomy with additional procedures | 11 (4.12) | 1 (1.12) | 10 (5.62) | |
| Laparoscopic inguinal hernia repair | 2 (0.75) | 1 (1.12) | 1 (0.56) | |
| Open inguinal hernia repair | 13 (0.75) | 4 (4.49) | 9 (5.06) | |
| Laparoscopic ventral or umbilical hernia repair | 9 (3.37) | 5 (5.62) | 4 (2.25) | |
| Open ventral or umbilical hernia repair | 21 (7.87) | 7 (7.87) | 14 (7.87) | |
| Incision and drainage (peri-rectal abscess) | 7 (2.62) | 3 (3.37) | 4 (2.25) | |
| Other | 2 (0.75) | 0 (0.00) | 2 (1.12) | |
| Any return to the OR, n (%)* | 9 (3.37) | 3 (3.37) | 6 (3.37) | 1.00 |
| Unexpected return to the OR, n (%)* | 8 | 3 | 5 | .45 |
| Inpatient procedures, n (%) | ||||
| Surgical chest tube placement | 1 (0.19) | 0 (0.00) | 1 (0.28) | .48 |
| Thoracentesis, with or without pigtail catheter placement | 6 (1.13) | 3 (1.69) | 3 (0.85) | .38 |
| Percutaneous cholecystostomy | 20 (3.77) | 6 (3.39) | 14 (3.95) | .75 |
| Percutaneous drainage procedure | 9 (1.69) | 1 (0.56) | 8 (2.26) | .15 |
| Upper endoscopy | 26 (4.90) | 8 (4.52) | 18 (5.08) | .78 |
| Colonoscopy | 2 (0.38) | 0 (0.00) | 2 (0.56) | .32 |
| Angiography with embolization or other procedure | 5 (0.94) | 4 (2.26) | 1 (0.28) | .03 |
| Other procedure† | 48 (9.04) | 11 (6.21) | 37 (10.45) | .11 |
| Total LOS, d, median [IQR] | 2.44 [1.51–4.80] | 2.82 [1.64–5.41] | 2.34 [1.46–4.66] | .13 |
| Hospital discharge disposition, n (%) | ||||
| Home or self-care | 427 (80.41) | 107 (60.45) | 320 (90.40) | <.001 |
| Inpatient rehabilitation facility | 13 (2.45) | 7 (3.95) | 6 (1.69) | |
| Skilled nursing facility | 13 (2.45) | 5 (2.82) | 8 (2.26) | |
| Long-term acute care | 4 (0.75) | 1 (0.56) | 3 (0.85) | |
| Discharged against medical advice | 17 (3.20) | 13 (7.34) | 4 (1.13) | |
| Inpatient psychiatric hospital or facility | 1 (0.19) | 1 (0.56) | 0 (0.00) | |
| In-hospital death | 8 (1.51) | 4 (2.26) | 4 (1.13) | |
| Shelter | 32 (6.03) | 32 (18.08) | 0 (0.00) | |
| Other | 16 (3.01) | 7 (3.95) | 9 (2.54) |
IQR, interquartile range; LOS, length of stay; OR, operating room.
The denominator for these outcomes includes only patients who underwent an operation or procedure during their hospitalization.
Includes paracentesis, central venous or hemodialysis catheter, bedside incision and drainage, MRCP (magnetic resonance cholangiopancreatography)/ERCP (endoscopic retrograde cholangiopancreatography), and cardiac catheterization with or without angioplasty.
Unhoused patients comprised 33.33% (n = 177) of the cohort, whereas housed patients represented 66.67% (n = 354). Unhoused patients were less likely to be privately insured; there were no significant differences in all other demographic variables between housed and unhoused patients in the matched cohort (Table I). There were no significant differences in types of surgical intervention between housed and unhoused patients. Rates of nonoperative procedures were similar between the groups, with the exception of noncoronary angiography, which occurred more often in unhoused patients (Table II). Unhoused patients were less likely to be discharged home (Table II). Specifically evaluating demographic, clinical, and outcome variables within the subset of patients who underwent a surgical intervention during hospitalization, results were similar to that of the overall cohort (Supplementary Table S3).
Unadjusted analysis
Overall, there were no significant difference in in-hospital adverse event rates between housed (11.3%) and unhoused (9.6%) patients (P = .55). Compared with housed patients, unhoused individuals had greater rates of stroke (1.13% vs 0.00%, P = .05) and lower rates of hemorrhage (0.00% vs 0.56%, P = .03); however, the incidence of these complications was quite low (Table III).
Table III.
Outcomes
| Characteristics | All patients | Unhoused | Housed | P value |
|---|---|---|---|---|
| (N = 531) | (n = 177) | (n = 354) | ||
| Complication during admission, n (%) | ||||
| Surgical-site infection or wound disruption* | 10 (1.88) | 3 (1.69) | 7 (1.98) | .82 |
| Pneumonia | 5 (0.94) | 1 (0.56) | 4 (1.13) | .53 |
| Unplanned intubation | 11 (2.07) | 3 (1.69) | 8 (2.26) | .67 |
| Pulmonary embolism | 4 (0.75) | 0 (0.00) | 4 (1.13) | .16 |
| Mechanical ventilation >48 h | 7 (1.32) | 2 (1.13) | 5 (1.41) | .79 |
| Acute renal failure | 18 (3.39) | 5 (2.82) | 13 (3.67) | .61 |
| Urinary tract infection | 13 (2.45) | 6 (3.39) | 7 (1.98) | .32 |
| Stroke/cerebrovascular accident | 2 (0.38) | 2 (1.13) | 0 (0.00) | .05 |
| Cardiac arrest | 3 (0.56) | 1 (0.56) | 2 (0.56) | 1.00 |
| Myocardial infarction | 1 (0.19) | 0 (0.00) | 1 (0.28) | .48 |
| Hemorrhage requiring transfusion | 2 (0.38) | 0 (0.00) | 2 (0.56) | .03 |
| Deep venous thrombosis | 9 (1.69) | 1 (0.56) | 8 (2.26) | .15 |
| Sepsis/septic shock | 15 (2.82) | 6 (3.39) | 9 (2.54) | .58 |
| Death | 6 (1.13) | 3 (1.69) | 3 (0.85) | .38 |
| ICU admission, n (%) | 19 (3.58) | 7 (3.95) | 12 (3.39) | .74 |
| Extended LOS, n (%) | 132 (24.86) | 50 (28.25) | 82 (23.16) | .20 |
| Follow-up appointment, n (%) | ||||
| With surgeon or subspecialist | ||||
| Scheduled | 317 (59.70) | 97 (54.80) | 220 (62.15) | .10 |
| Completed | 246 (46.33) | 62 (35.03) | 184 (51.98) | <.001 |
| With primary care provider | ||||
| Scheduled | 195 (36.72) | 75 (42.37) | 120 (33.90) | .06 |
| Completed | 141 (26.55) | 53 (29.94) | 88 (24.86) | .21 |
| No dedicated follow-up appointment | 183 (34.46) | 76 (42.94) | 107 (30.23) | .004 |
| Return to ED within 30 d, n (%) | 144 (27.22) | 74 (42.05) | 70 (19.83) | <.001 |
| Related to initial injury/hospitalization | 91 (63.19) | 40 (54.05) | 51 (72.86) | .02 |
| Readmission within 30 d, n (%) | 70 (13.21) | 32 (18.08) | 38 (10.76) | .02 |
| Related to initial injury/diagnosis | 50 (71.43) | 20 (62.50) | 30 (78.95) | .13 |
| Operative intervention during readmission | 6 (8.57) | 1 (3.12) | 5 (13.16) | .14 |
| Principal reason for readmission, n (%) | ||||
| Deep space infection | 3 (4.29) | 1 (3.12) | 2 (5.26) | .14 |
| Bleeding | 2 (2.86) | 2 (6.25) | 0 (0.00) | |
| SIRS/sepsis | 3 (4.29) | 0 (0.00) | 3 (7.89) | |
| Pain | 27 (38.57) | 12 (37.50) | 15 (39.47) | |
| Wound dehiscence | 1 (1.43) | 1 (3.12) | 0 (0.00) | |
| Social issues | 1 (1.43) | 0 (0.00) | 1 (2.63) | |
| Cardiac event | 2 (2.86) | 1 (3.12) | 1 (3.12) | |
| Related to chronic condition (ie, COPD) | 7 (10.00) | 1 (3.12) | 6 (15.79) | |
| Failure to thrive | 4 (5.71) | 1 (3.12) | 3 (7.89) | |
| Other | 20 (28.57) | 13 (40.62) | 7 (18.42) |
COPD, chronic obstructive pulmonary disease; ED, emergency department; ICU, intensive care unit; LOS, length of stay; SIRS, systemic inflammatory response syndrome.
The denominator for these outcomes includes only patients who underwent an operation or procedure during their hospitalization.
There was no significant difference in ICU admission or extended length of stay between the groups (Table III). Unhoused patients were significantly less likely to be scheduled for and to attend surgery clinic follow-up, but there were no differences in rates of primary care follow-up attendance (Table III). Unhoused patients were more likely to have no dedicated follow-up within 30 days of discharge (42.94% vs 30.23%, P = .004) (Table III). Unhoused patients returned to the emergency department at a rate of 42.05% within 30 days of hospital discharge, compared with 19.83% in housed patients (P < .001); however, they were less likely to return for a chief complaint related to the index hospitalization (Table III). Similarly, unhoused patients were more likely to have an unplanned readmission within 30 days of discharge (18.08% vs 10.76%, P = .02), but there were no differences between housed and unhoused patients in reason for readmission, relationship of readmission to index admission, or need for operative intervention during readmission (Table III). On sensitivity analysis, patients discharged home had greater follow-up attendance, lower emergency department use, and lower readmission rates (Supplementary Table S4).
Adjusted analysis
After we controlled for age, sex, primary language, insurance status, housing status, Charlson Comorbidity Index, diagnosis group, surgical intervention (dichotomous), and ICU admission, there were no differences in in-hospital complication rates between housed and unhoused patients (odds ratio [OR], 0.76; 95% confidence interval [CI], 0.35–1.69, P = 0.51) (Figure 2). Unhoused patients had no differences in rates of ICU admission or extended length of stay (Figure 2). Unhoused patients had lower odds of completing outpatient follow-up (OR, 0.54; 95% CI, 0.35e0.85, P = .008). Within 30 days of hospital discharge, unhoused patients had significantly greater odds of emergency department use (OR, 2.72; 95% CI, 1.78–4.14, P < .001) and unplanned readmissions (OR, 1.87; 95% CI, 1.10–3.19, P = .02) than their housed counterparts (Figure 2).
Figure 2.
Adjusted analysis: association between housing status and outcomes multivariable conditional logistic regression analysis with outcome as listed, and predictors including housing status (dichotomous), age (continuous), sex (dichotomous), English as primary language (dichotomous), commercial insurance (dichotomous), diagnosis group (categorical), Charlson Comorbidity Index (continuous), surgical intervention (dichotomous), and ICU admission/transfer (when appropriate, dichotomous). ORs are presented on a logarithmic scale, with diamonds representing OR and error bars representing 95% CI. CI, confidence interval; ED, emergency department; ICU, intensive care unit; OR, odds ratio.
Sensitivity analysis including only surgical patients (n = 267) found no differences in in-hospital complication rates or extended length of stay between housed and unhoused patients, although ICU admission was unable to be evaluated because of small sample size (Supplementary Table S3). Unlike in the overall cohort, there was no significant difference in 30-day follow-up completion; however, unhoused patients in this subgroup continued to demonstrate greater odds of emergency department utilization and unplanned readmission within 30 days.
Discussion
In this single-institution matched cohort study evaluating the outcomes of unhoused patients admitted for emergency general surgery diagnoses, there were no significant differences in in-hospital outcomes, including complications, ICU admission rates, or extended length of stay, contrary to our hypothesis. However, consistent with our hypothesis, unhoused patients had significantly lower odds of attending follow-up appointments and greater odds of emergency department use and unplanned readmission within 30 days of discharge.
Our results with respect to in-hospital outcomes are somewhat in contrast with previous studies of unhoused patients, which have demonstrated significant delays to surgical care5,9 and more emergent surgical needs,20 which would portend a greater complication risk. Contrary to previous literature, we did not find any difference in extended length of stay or ICU admissions between housed and unhoused patients.10,21 These discrepancies may be related to the focus solely on patients who undergo emergency general surgery, as the emergent nature of the complaint may eliminate some disparities related to delays in seeking urgent or emergent care between housed and unhoused patients.
We found that unhoused patients had lower odds of follow-up clinic visit attendance and greater odds of emergency department visits and unplanned readmissions. These findings are consistent with previous studies, which have found greater rates of all-cause emergency department visits7,11,12,22 and readmissions12,13 in unhoused patients. Similar findings have been demonstrated in patients who undergo surgery12,13 but not specifically in the emergency general surgery population. As unhoused patients were less likely to have any follow-up appointment made, this may suggest that providers may be modifying management based on housing status. Although there were lower clinic follow-up rates in unhoused patients, we only included clinic visits that were directly related to the index admission, which may skew our findings. In addition, although sample size limited the granularity of our regression, it is possible that provider specialty impacted rates of clinic attendance, as our unadjusted analysis demonstrated significant differences in surgical clinic visit rates but not in primary care visits. Of note, our sensitivity analysis of only patients who underwent an operation found no difference in follow-up attendance but continued to demonstrate greater odds of emergency department visits and readmissions, which supports previous data showing that outpatient follow-up attendance is not protective against these outcomes.20 Overall, our findings of lower odds of follow-up attendance and greater odds of emergency department visits and readmissions in these patients suggest a need for more tailored follow-up and health care system engagement for this patient population experiencing poor health care access.
Overall, our findings support that disparities in posthospitalization care are present for unhoused patients who undergo emergency general surgery. Significant disparities in access to care are well-documented in the unhoused patient population, an already marginalized group with high rates of unmet health needs and multifaced barriers to care.4,9 Up to 13% of unhoused patients report difficulty obtaining primary care at safety net facilities,23 and 16% reporting being rejected by the care locations they seek out.24 More than one half of unhoused patients self-report receiving the bulk of their medical care through the emergency department,23 whereas 32% have difficulty knowing where to seek care other than the emergency department.24 In our study, unhoused patients had greater rates of using the emergency department for reasons unrelated to their index admission, supporting the idea that this patient population may use the emergency health care system for all health needs. Although we were unable to evaluate the association between emergency department visits and subsequent readmissions, previous literature on acute care surgery has found an approximately 8% emergency department visit rate within 30 days, with 36.5% of emergency department visits resulting in a readmission.25
Intervention-based studies are relatively sparse; however, it has been shown that comprehensive care provision tailored specifically to the needs of the unhoused population may improve outcomes.24 Access to community clinics, community health workers, and nursing may improve postdischarge health status.24,26–30 Specifically, initiatives such as “street medicine,” involving meeting patients in the community to provide care may help to mitigate some access disparities.27 Ideally, these programs should be collaborative with community health care facilities and hospitals and provide comprehensive programs to include social support.27 In the emergency general surgery patient population, more research is needed to optimize access to posthospitalization care, with the downstream goal of reducing unnecessary emergency department use and curbing high readmission rates in unhoused patients.
Study limitations
This study has several limitations. As a single-institution retrospective study, the generalizability of our results is limited, and we cannot determine causality between housing status and our outcomes of interest. The relatively small sample size and matching algorithm precludes analysis of all possible confounders and mediators, which should be considered in future larger-scale studies. Matching and our exclusion criteria may also create a sample that is not entirely representative of the overall unhoused patient population. Although we employed a multistep process to determine housing status, limitations in documentation in the electronic medical record may result in potential misclassification of our primary exposure. Although our in-hospital outcomes were similar between groups, we did not evaluate the urgency of operations performed. Given sample size, our study may not be adequately powered to determine differences in in-hospital outcomes. In addition, we were only able to capture instances of care (including follow-up, emergency department visits, and readmissions) that occurred at our own institution. This may be remedied in future studies through multi-institutional collaboration. Finally, the use of a 30-day period for outcome measures, although standard in the literature, may not capture the full extent of health care system contact related to the patient’s surgical diagnosis.
In conclusion, in this single-institution study of patients who undergo emergency general surgery at an urban safety net hospital, unhoused patients had similar in-hospital complication rates. However, they had lower follow-up rates and greater odds of emergency department use and unplanned readmission, indicating a need for access to posthospitalization resources to improve outcomes. This requires a collaborative effort between health systems, public health, and the social services sector to focus on community development, addressing social determinants of health to improve outcomes.
Supplementary Material
Acknowledgments
The authors thank Florencia A. Pereira, MD, Lucero Paredes, MD, and Javier Galindo, MD, for their contributions to data collection.
Funding/Support
S.M.S. and A.J.K.H. report receiving partial salary support from a National Institutes of Health T32 Training Grant (GM086308).
Footnotes
CRediT authorship contribution statement
Sophia M. Smith: Writing – review & editing, Writing – original draft, Project administration, Methodology, Formal analysis, Data curation. Brendin R. Beaulieu-Jones: Writing – review & editing, Writing – original draft, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. Maia R. Nofal: Writing – review & editing, Methodology, Data curation. Anna J. Kobzeva-Herzog: Writing – review & editing, Methodology, Data curation. Emily J. Ha: Writing – review & editing, Data curation. Heejoo Kang: Writing – review & editing, Data curation. Tracey A. Dechert: Writing – review & editing, Methodology, Investigation. Sabrina E. Sanchez: Writing – review & editing, Supervision, Project administration, Methodology, Formal analysis, Conceptualization. Megan G. Janeway: Writing – review & editing, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.
Supplementary materials
Supplementary material associated with this article can be found, in the online version, at [https://doi.org/10.1016/j.surg.2024.08.012].
Conflict of Interest/Disclosure
The authors declare that they have no competing interests.
References
- 1.de Sousa T, Andrichik A, Cuellar M, Marson J, Prestera E, Rush K. The 2022 Annual Homelessness Assessment Report (AHAR to Congress) part 1: point-in-time estimates of Homelessness, December 2022. The U.S. Department of Housing and Urban Development Office of Community Planning and Development; 2022. https://www.huduser.gov/portal/sites/default/files/pdf/2022-ahar-part-1.pdf. Accessed March 20, 2024. [Google Scholar]
- 2.Sandel M, Desmond M. Investing in housing for health improves both mission and margin. JAMA. 2017;318:2291e2292. [DOI] [PubMed] [Google Scholar]
- 3.Kushel MB. Factors associated with the health care utilization of homeless persons. JAMA. 2001;285:200. [DOI] [PubMed] [Google Scholar]
- 4.Vuillermoz C, Vandentorren S, Brondeel R, Chauvin P. Unmet healthcare needs in homeless women with children in the greater Paris area in France. PLoS One. 2017;12:e0184138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Baggett TP, O’Connell JJ, Singer DE, Rigotti NA. The unmet health care needs of homeless adults: a national study. Am J Public Health. 2010;100:1326–1333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fazel S, Geddes JR, Kushel M. The health of homeless people in high-income countries: descriptive epidemiology, health consequences, and clinical and policy recommendations. Lancet. 2014;384:1529–1540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Amato S, Nobay F, Amato DP, Abar B, Adler D. Sick and unsheltered: homelessness as a major risk factor for emergency care utilization. Am J Emerg Med. 2019;37:415–420. [DOI] [PubMed] [Google Scholar]
- 8.Pearson DA, Bruggman AR, Haukoos JS. Out-of-hospital and emergency department utilization by adult homeless patients. Ann Emerg Med. 2007;50: 646–652. [DOI] [PubMed] [Google Scholar]
- 9.Festa K, Hirsch AE, Cassidy MR, et al. Breast cancer treatment delays at an urban safety net hospital among women experiencing homelessness. J Community Health. 2020;45:452–457. [DOI] [PubMed] [Google Scholar]
- 10.Bigé N, Hejblum G, Baudel JL, et al. Homeless patients in the ICU: an observational propensity-matched cohort study. Crit Care Med. 2015;43:1246–1254. [DOI] [PubMed] [Google Scholar]
- 11.Bowen M, Marwick S, Marshall T, et al. Multimorbidity and emergency department visits by a homeless population: a database study in specialist general practice. Br J Gen Pract. 2019;69:e515ee525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Decker HC, Kanzaria HK, Evans J, Pierce L, Wick EC. Association of housing status with types of operations and postoperative health care utilization. Ann Surg. 2023;278:883–889. [DOI] [PubMed] [Google Scholar]
- 13.Titan A, Graham L, Rosen A, et al. Homeless status, postdischarge health care utilization, and readmission after surgery. Med Care. 2018;56:460–469. [DOI] [PubMed] [Google Scholar]
- 14.DePolo N, Bathan K, Tran Z, et al. Prospective assessment of social determinants of health and length of stay among emergency general surgery and trauma patients. Am Surg. 2023;89:4186–4190. [DOI] [PubMed] [Google Scholar]
- 15.Hircock C, Huan P, Pizzola C, McDonald M. A scoping review of surgical care for people experiencing homelessness: prevalence, access, and disparities. Can J Surg. 2024;67:E27eE39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inf. 2009;42:377–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inf. 2019;95:103208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Krell RW, Girotti ME, Dimick JB. Extended length of stay after surgery: complications, inefficient practice, or sick patients? JAMA Surg. 2014;149:815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee DC, Liang H, Shi L. The convergence of racial and income disparities in health insurance coverage in the United States. Int J Equity Health. 2021;20:96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Smith SM, Zhao X, Kenzik K, Michael C, Jenkins K, Sanchez SE. Scheduled follow-up and association with emergency department use and readmission after trauma. J Am Coll Surg. 2024;239:234–241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hwang SW, Aubry T. Hospital costs and length of stay among homeless patients admitted to medical, surgical, and psychiatric services. Med Care. 2011;49:350–354. [DOI] [PubMed] [Google Scholar]
- 22.Burak A, Wyszkowska Z, Liebert A, Kłopocka M. Profile of homeless people in emergency departments. Int J Occup Med Environ Health. 2022;35:157–167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kertesz SG, McNeil W, Cash JJ, et al. Unmet need for medical care and safety net accessibility among Birmingham’s homeless. J Urban Health. 2014;91:33–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hwang SW, Burns T. Health interventions for people who are homeless. Lancet. 2014;384:1541–1547. [DOI] [PubMed] [Google Scholar]
- 25.Brown CS, Yang J, Meng Z, Henderson J, Dimick JB, Telem DA. Trends in emergency department utilization following common operations in New York State, 2005e2014. Surg Endosc. 2020;34:1994–1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kertesz SG, Posner MA, O’Connell JJ, et al. Post-hospital medical respite care and hospital readmission of homeless persons. J Prev Interv Community. 2009;37:129–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Howe EC, Buck DS, Withers J. Delivering health care on the streets: challenges and opportunities for quality management. Qual Manag Health Care. 2009;18: 239–246. [DOI] [PubMed] [Google Scholar]
- 28.Whelan C, Chambers C, Chan M, Thomas S, Ramos G, Hwang SW. Why do homeless people use a mobile health unit in a country with universal health care? J Prim Care Community Health. 2010;1:78–82. [DOI] [PubMed] [Google Scholar]
- 29.Nyamathi A, Kennedy B, Branson C, et al. Impact of nursing intervention on improving HIV, hepatitis knowledge and mental health among homeless young adults. Community Ment Health J. 2013;49:178–184. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Carter J, Hassan S, Walton A, Yu L, Donelan K, Thorndike AN. Effect of community health workers on 30-day hospital readmissions in an accountable care organization population: a randomized clinical trial. JAMA Netw Open. 2021;4:e2110936. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.